The goals of this project are to continue our investigation into statistical methods to assess the effects of covariates in longitudinal studies. Such studies arise in a wide variety of medical investigations including clinical trials, observational studies, and multistate survival studies. Our primary area of application is to the treatment- recovery process following a bone marrow transplant, but the techniques are applicable to a wide range of medical studies. Our specific goals are three-fold. First, we plan to study techniques for modeling time varying effects of fixed covariates; to study the effects of measurement error on predicting patient outcome in survival experiments and to study inference for survival studies based on non- cohort based sampling schemes. Second, we plan to study multivariate methods in survival analysis including dynamic modeling of patient prognosis, dynamic modeling of causal effects in survival analysis and models for incorporating association within groups of patients into an analysis. We also plan a series of prototypical detailed studies of the bone marrow transplantation recovery process using data from the International Bone Marrow Transplant Registry and the Autologous Blood and Marrow Transplant Registry. Third, we plan to investigate statistical methods for longitudinal data where one has repeated measurements of a common event or periodic observation of discrete and continuous responses. We will develop statistical software for the techniques we develop and apply these methods to medical data available to us through our collaborations with medical investigators.
Showing the most recent 10 out of 74 publications
